Path dependence in hierarchical organizations: The ... · PDF filePath dependence in hierarchical organizations: The influence of environmental ... technological change on the TV market

Path dependence in hierarchical organizations:

The influence of environmental dynamics

Arne Petermann

Institute for Quality and Management, BAGSS

Konrad-Zuse-Str 3a, 66115 Saarbrücken, Germany

E-mail: [email protected]

KEYWORDS

agent based modelling, social simulation, path

dependence, path breaking, organizational studies,

business strategy, technology strategy

ABSTRACT

The following paper will describe how path dependent

hierarchical organizations are affected by a changing

environment. The results of current research in this field

(Petermann et al. 2012) analyzed path dependency of

norms and institutions in different kinds of hierarchical

organizations and the impact of leadership within this

process. The results were produced for stable

environments only. Agent based simulation was applied

as research method. In order to examine how this

process evolves when the organizational environment is

changing, the existing model will be enhanced. The

objective is to simulate the impact of external influences

to the emergence of norms within an organization.

INTRODUCTION

Nowadays most organizations have to deal with a

changing environment. From the organizational point of

view a changing environment can be seen as

disturbances from outside, that forces the organization

to adapt. If an organization fails to do so, it may fall

back or even be eliminated from the competition. This

comes with a high risk, especially when new

technologies flood the market and companies have to

react. Examples may be found by taking a closer look at

companies like Loewe or Nokia. Loewe missed the

technological change on the TV market from the CRT

displays to the new LCD-based flat screen technologies.

In fact, Loewe still builds CRT displays. The result is an

imbalance of supply and demand, because most

customers are not interested in those TV’s any more.

Thus Loewe appears ignorant of market realities. The

high technical level of their obsolete skills is disguising

the internal view of the environment, in this case:

innovations on the TV market. In the end the investor

Stargate Capital bought Loewe and made some serious

changes. But their previous ignorance almost led them

into bankruptcy.

Nokia on the other hand was one of the pioneer

companies on the mobile phone market, but they did not

Alexander Simon

Berlin University of Professional Studies

Katharinenstraße 17-18, 10711 Berlin, Germany

E-mail: [email protected]

react adequately to new mobile trends. Just like Loewe,

Nokia suffered immensely when other suppliers like

Apple and Samsung captured the market applying the

latest technologies. By now the mobile phone division

of Nokia has been bought by Microsoft. The questions

that arise are: why do companies sometimes need to get

hit so hard from external influences until they see that

they have to change? How fierce do these influences

need to be?

In the following research the model M1, (Petermann et

al. 2012) which for reasons of simplicity was built on

the assumption of a stable environment will be extended

with a new variable one or the other will include

environmental change into the model.

LITERATURE REVIEW AND RESEARCH

QUESTION

The theoretical concept for the behavioral analysis

described above is called “theory of path dependence”.

The concept of that theory was first described by David

(1985). He dug into the history of the “QWERTY”-

keyboards from their first steps in the 19th century until

1985. This alignment of characters has been dominant

till today for nearly 100 years. In the early 1930s the

alternative “DVORAK” keyboard layout was

developed. In that time this new technology was clearly

a superior solution than the incumbent. These

keyboards, however, were not able to become a serious

competitor to “QWERTY”- keyboards. David examined

in detail the self-reinforcing mechanisms that led to the

domination of the established keyboards by the inferior

QWERTY solution.

Based on David's findings, Arthur (1989) used a polya

urn model to analyze the self reinforcing mechanisms

discovered by David in a more formal way. In his model

two technologies (A and B) are entering the market at

the same time and compete for the adoption by

customers called agents. At the beginning both

technologies have the same chances to get adopted. For

the first time in the history of the path dependence

debate, Arthur coined the definition of the historical

small events increasing returns and contingency. These

events are responsible for the start of the path process

and lead to a lock-in of the technologies A or B. Figure

1(Arthur 1989: 120) illustrates this behavior. When B is

locked in, A is completely eliminated from the market.

Proceedings 30th European Conference on Modelling and Simulation ©ECMS Thorsten Claus, Frank Herrmann, Michael Manitz, Oliver Rose (Editors) ISBN: 978-0-9932440-2-5 / ISBN: 978-0-9932440-3-2 (CD)

mailto:[email protected]

http://www.dict.cc/english-german/imbalance+of+supply+and+demand.html

mailto:[email protected]

Figure 1: Increasing returns adoption: a random walk

with absorbing barriers

David and Arthur stress that a technology can become

dominant even when it is inferior in terms of its long

term value to the system.

Transform path dependence to an organizational

context

To transfer the theory of path dependence to an

organizational context, a different view of Arthur’s

description is needed. In organizations and social

systems history always matters, and due to the ongoing

variations in behavior, the lock-in on markets has

peculiar characteristics. There is less adoption behavior;

hence development phases deviate from purely

technological path dependence. To capture

organizational path dependence, Sydow developed a

model which describes this advanced concept of path

dependence. In this model the path is split into three

different phases. Figure 2 (Sydow et al., 2009: 692)

shows the concept of this model.

Figure 2: The Constitution of an Organizational Path

Phase 1: Preformation Phase

In this phase the decisions the participants are able to

make are relatively open. Influences at this point can be

historical events, “history matters”, routines, and the

existing culture of the organization. In the beginning the

participants already have an idea of thinking and

behaving in their daily environments. Koch (2007: 286)

described imprinting circumstances of an organizational

culture in this context. Therefore the decisions that will

be made in the future are already not completely open.

In figure 2 all options are symbolized by the black stars,

but only the stars in the grey zone are available options

for the organization.

Phase 2: Formation Phase

In this phase the path begins to emerge. The step from

phase 1 to phase 2 is called “critical juncture”. An

unknown or virtually unrecognizable event from the

past leads to the organizational path formation (Sydow

et al., 2009: 693). These events are described as “small

events”. The self-reinforcing effects that are triggered

by these small events narrow the path and limit future

choices within the organization.

Phase3: Lock-In Phase

The reinforcing effects have now taken the lead and

reduced the scope of choices drastically. The

organizational path has become locked-in. The lock-in

state may, in an unfortunate case, be an inefficient one

which disables the organization's ability to change and

adopt more efficient solutions to the problems at hand.

As described, Loewe appeared locked-in to such an

unfortunate state. At first the state was very efficient,

but when the market changed Loewe’s technology was

not needed anymore and thus the state lost its efficiency,

causing severe problems for the organization.

RESEARCH QUESTIONS

We are interested in the impact of a changing

environment to organizations that undergo path

depended processes. In historical analysis scholars have

shown many examples of organizations that were able

to adapt in the light of changing environment, while

others are stuck in a lock-in state, unable to change,

even when the necessity to change became obvious

from an outside perspective. Our model aims at

describing hierarchical organizations, that undergo a

path depended development in a changing environment.

Will they be able to adapt or do they stick to the path?

What can we learn about this process applying

simulation methods? How should an organization be

structured, to be able to adapt in the light of dramatic

changes in the environment?

METHOD

In modern social and management sciences the method

of simulation modeling has been accepted since the

early 1990s (Harrison 2007: 1232). When complexity

and non-linearity of social systems make it hard or

impossible to develop mathematical equations,

simulation models are a good choice to describe the

whole system and its development (Gilbert et al 2005:

16). ‘Simulation is particularly useful when the

theoretical focus is longitudinal, nonlinear, or

processual, or when empirical data are challenging to

obtain’ (Davis et al, 2007: 481). On the other hand, it is

important to know that the method of simulation cannot

replace empirical or analytic methods, but it can provide

insights and first assumptions for other social research

methods.

The basic model

The basics of this research is the simulation model M1

Petermann et al. (2012) developed in their simulation

study about the competing powers of self-reinforcing

dynamics and hierarchy in organizations. The theory of

that model is the simulation of a norm A and a norm B

in an organizational hierarchy structure and to answer

the question which norm will be adopted by most of its

members. Every member of the organization is

represented as an agent. These agents are able to decide

whether to adopt norm A or norm B.

Agents decision algorithm to adopt A or B

To implement this technically, the agents need to be

forced to adopt a norm. Therefore the force-to-act

variable FTA is defined (Petermann et al. 2012: 726).

(1)

Vj describes the connection of individual and

organizational goals according to Vrooms (1964)

expectancy theory. Ej ϵ [0,1] represents the subjective

probability of each agent’s decision. This variable

represents the “small events” of the organizational path

dependence theory. To implement this in the algorithm,

the strictly monotonously increasing function

(2)

is used in the simulation to determine V according to

equation (1) with M ϵ {A, B}, m = 1 for fA,c(x) and m=

-1 for fB,c(x). The variable c represents the reinforcing

effects and is generated by the actual spread ϵ [-1, 1]

which is a variable that characterizes the state of the

system, which is either dominated by agents who all

choose A (spread =-1) or agents who all choose B

(spread=1) or at some state between these extreme cases

(spread between -1 and 1). The factor i(y) sets the value

of li in the correction path direction. This could be 1 or

-1. At the beginning of the simulation the spread is 0

(meaning there are equally large groups of agents

choosing A and B in the beginning of the simulation).

The lock-in state is nearly 1 for A or nearly -1 for B

after a defined amount of time (measured in ticks). The

misfit costs are described by x. The leadership impact

variable li, which makes the simulation of a hierarchy

organizational structure possible, is affecting every

agent according to what norm his supervisor prefers.

Under these conditions the agents choose an adoption

for A, when

(3)

and otherwise B if FTAB > FTAA.

Simulation of an external impact

Enhancing this model further, we now implement an

external impact into the FTA function to see whether or

not this will have an effect of breaking the

organizational path. Therefore, equation (2) needs to be

extended with an additional value.

(4)

The variable ei represents the external impact from the

changing environment. The factor s(z) is only used to

set the correct direction, which depends on the actual

path. The value generation of that variable, needs to be

clarified in the next step. While all other variables in the

equation are generated by the simulated organization

itself, ei is triggered from an external source. When

there is no external impact, ei is equal to 1 and behaves

neutrally. The question of how the model reacts after the

lock-in, has occurred is highly interesting. Are there any

options to “reset” the norm distribution of the

organization? The goal here is to find out about the

behavior of the organization regarding the external

impact. Is its intensity, its continuity, or a mix of both

able to break the path? Every agent in the system is

subject to the same external impacts. We assume that

environmental influences have the same strength

throughout the organization.

SIMULATION

To simulate the described external impact, we need to

specify in what way and when the variable ei should

change. The first condition, we need to break the path

and the path must be locked-in. That means that a

dominant norm exists in the simulation model. The

lock-in state in model M1 is defined by Petermann et al.

(2012: 195) as minimum 500 of ticks with a spread >

0.9 if B is dominant (spread < -0.9 if A is dominant).

Furthermore, a definition for breaking a path is also

needed. The model M1 defines no values for that, so we

assume a path is broken when spread < 0.5, when B was

the current norm in the company and a spread > -0.5

when A was the current norm. This means that less than

75% of all company members adopted a norm. The last

variable is the leadership impact. This is set to 1, to

have an impact from that side. The defined values for

lock-in and leadership impact are assumption and not

empirical researched values.

The variation of the external impact is possible in two

ways. Either the intensity can be variated or the amount

of time (number of ticks) the impact is present in the

system. To get usable data from the simulation model,

only datasets with a lock-in at B before the external

impact is triggered are used for analysis purposes.

Otherwise it is not possible to see a behavior for one

norm. A simulation for each parameter-set will run 100

times according to the Monte-Carlo method (Law et al.

1991:113).

Run of the 1. Simulation

The change of the external impacts must be further

clarified to run the first approach. During the

enhancement of the model the following parameters

seemed to be valid for a first run. After the first

simulation an optimization of the parameters is needed.

Maybe a closer look at several parameter areas is

interesting.

intensity (int)

con

tin

uit

y (

tick

s)

1 3 5 7 10

10 10/1 10/3 10/5 10/7 10/10

40 40/1 40/3 40/5 40/7 40/10

70 70/1 70/3 70/5 70/7 70/10

100 100/1 100/3 100/5 100/7 100/10

130 130/1 130/3 130/5 130/7 130/10

Figure 3: 1. Simulation parameter Setup

The lock-in behavior with leadership impact of 1 and 2

is at 6000 ticks (Petermann et al. 2012:195). This

means, that each simulation must run at least 6000 ticks,

before the external impact can be triggered. A complete

run will last 8500 ticks, and then the system has enough

time to reconfigure itself after the external impact. It is

not possible to define a number of ticks after the impact,

when the system has locked-in again. This basic setup is

used for all simulations in this paper; otherwise it is not

possible to compare the results properly.

Results of the 1st Simulation

The result of this run is a huge amount of data, which

needs to be analyzed. The first intensity parameters

from 1 to 3 will not be visualized in this paper, because

the maximum possible spread change is from 1 to 0,992

at a random point of time, so with a parameter

combination of 130 for continuity and 3 for intensity no

connection to the external impact can be identified. The

effects that occur at the intensity of 10 are also not

visualized they are similar to the graphics that depict the

intensity of 7.

Results for Intensity of 1 and 3

No valid differences could be detected, that change the

system normal behavior. It is not possible, to force a

path break with all combinations containing the

parameter 1 for intensity.

The most interesting effect occurs at the parameter

intensity between 5 and 7.

Results for Intensity of 5

Figure 4: Left: Histograpchical view: probability

density, 150 ticks summarized. X axis: spread, Y axis:

probability from 0-1, z-Axis: time in ticks, starts

counting at 5800 ticks. Right: exemplary first 10 runs. X

Axis: ticks, Y-Axis: spread.

At this parameter setup the system started to react. With

the combination of continuity of 10 until continuity of

70 no valuable reactions are noticeable. However, at a

continuity of 100 the system starts to change. The

spread is forced to the path breaking direction. Of

course, it is only a spread of 0.992, but the events occur

exactly at the starting point of the external impact.

With this first result it is maybe useful, to increase the

continuity over 130 with an intensity of 5 to generate a

path break.


F






An intensity of 7 forces the system to break the path for

the first time even with low continuities. With a

continuity of 10 a significant system behavior is

detected, but there is still no path break. This happens

for the first time with a continuity of 40 (spread < 0.4).

Higher continuities with values of 70, 100 and 130

forced the system to establish new norms.


The intensity of 10 behaves nearly like the intensity of

7. With higher amount of continuity path breaks and

new path formations are results of the simulation.

Summary of the 1st Simulation

The first simulation run gave first insights to the system

behavior of the described model M1. In the following

three figures all parameter combinations described in

figure 3 are being compared.

The effect at the path is listed in figure 6. As described

above, the interesting area is between the intensities of 5

and 7. The probability of a path breaking behavior

increases rapidly at this interval. This parameter field

will be investigated more closely.

intensity (int)

con

tin

uit

y (

tick

s)

1 3 5 7 10

10 0% 0% 0% 0% 0%

40 0% 0% 0% 99% 100%

70 0% 0% 0% 100% 100%

100 0% 0% 0% 100% 100%

130 0% 0% 0% 100% 100%

Figure 6: Pathbreaking probability

Figure 7 describes the probability of new path

directions. With a continuity of 70 at an intensity of 7

the probability is 3% higher compared to an intensity of

10. There is, however, a small probability at 100

simulation runs that the result differs from one’s

expectations. That the system behaves unexpectedly at

this point could also be an assumption. A deeper

analysis about this could be an interesting question for

upcoming research, but it will not find place in this

paper.

intensity (int)

con

tin

uit

y (

tick

s)

1 3 5 7 10

10 0% 0% 0% 0% 0%

40 0% 0% 0% 0% 0%

70 0% 0% 0% 35% 32%

100 0% 0% 0% 100% 100%

130 0% 0% 0% 100% 100%

Figure 7: new path direction probability

Finally the average spread over all combinations is

shown in figure 8. The average was calculated at the last

tick of the impact. As expected, the spread changes in

the intensity fields of 1 and 3 are out of scope. It’s

interesting to see that with an intensity of 5 differs with

3%, but that is not according to its continuity.

intensity (int)

con

tin

uit

y (

tick

s)

1 3 5 7 10

10 1,00 1,00 1,00 0,84 0,82

40 1,00 1,00 1,00 0,40 0,37

70 1,00 1,00 0,98 0,05 0,02

100 1,00 1,00 0,97 -0,23 -0,25

130 1,00 1,00 1,00 -0,43 -0,45

Figure 8: Average spread, calculated at the last external

impact tick

The expected behavior was successfully created in the

first simulation run. The path was broken, and new path

directions emerged in the system. It is also a validation

of the external impact implementation of the model M1.

Furthermore interesting results came out of the first run.

A second and third simulation run are recommend at

this point. In the second simulation a closer look is

taken at the system behavior with very high intensities.

As described above, the intensities 7 and 10 seem to

behave almost equally. To clarify this interpretation, a

second simulation was done. The third simulation takes

a closer look at the intensity area between 5 and 7. Here

the system seems to react very sensitively.

Run of the 2nd Simulation

The finding of the first simulation: “the intensities of 7

and 10 behave almost equally” should be clarified in

this simulation. Therefore a simulation with highly

overdriven intensities was performed. We assume here

that there are no significant different system-behaviors

observables, with intensities of 10 or more. To break the

path with a continuity of 10 is also part of this

simulation. The path was influenced in the first

simulation with this continuity, but there was no

sustainable effect, like breaking or new formation, at the

path. Figure 9 shows the parameter setup for the second

simulation.

intensity (int)

con

tin

uit

y (

tick

s)

13 16 19

10 10/13 10/16 10/19

40 40/13 40/16 40/19

70 70/13 70/16 70/19

100 100/13 100/16 100/19

130 130/13 130/16 130/19

Figure 9: 2. Simulation parameter Setup

Results of the 2nd Simulation

The figures 10 – 12 show the visualized result of the

simulation.
















probability from -1, z-Axis: time in ticks, starts counting

at 5800 ticks. Right: exemplary first 10 runs. X Axis:

ticks, Y-Axis: spread.

The average spreads shall verify the assumption of a

constant changing behavior with increasing intensities.

The detailed results are shown in figure 13 to 15.

intensity (int)

con

tin

uit

y (

tick

s)

7 10 13 16 19

10 0% 0% 0% 0% 0%

40 99% 100% 100% 100% 100%

70 100% 100% 100% 100% 100%

100 100% 100% 100% 100% 100%

130 100% 100% 100% 100% 100%


intensity (int)

con

tin

uit

y (

tick

s)

7 10 13 16 19

10 0% 0% 0% 0% 0%

40 0% 0% 0% 0% 0%

70 35% 32% 49% 51% 60%

100 100% 100% 100% 100% 100%

130 100% 100% 100% 100% 100%


intensity (int)

con

tin

uit

y (

tick

s)

7 10 13 16 19

10 0,84 0,82 0,81 0,81 0,81

40 0,40 0,37 0,35 0,35 0,35

70 0,05 0,02 0,01 0 0

100 -0,23 -0,25 -0,25 -0,25 -0,26

130 -0,43 -0,45 -0,45 -0,45 -0,45

Figure 15: Average spread, calculated at the last

external impact tick

Figure 13 shows the pathbreaking probability for the

second simulation. All intensities have similar values,

except the parameter combination intensity of 7 and

continuity of 30 with 99%. The average spread, which

is shown in figure 15, only varies from the intensity

from 7 to 10. For an intensity of 13 or more the average

spread is constant. This result validates the assumption

that a continuity of 10 can not trigger a pathbreak in the

system, regardless the height of intensity. The most

interesting outcomes of this simulation are the values of

the new path direction probability (figure 14). The

continuities have the same behaviors, except 70. With a

higher intensity the new path direction probability

increases. Even if the average spread is 0 at a continuity

of 70 and intensity of 16 and continuity of 70 and

intensity of 19, that means a total balance between the

adopted norms A and B is present, the new path

direction probability increases about 9% between this

two combinations.

Run of the 3rd simulation

This simulation researches the area of the intensities

between 5 and 7 as described above. The following

table shows the parameter setup for this run.

intensity (int)

con

tin

uit

y (

tick

s)

5,5 6 6,5

10 10/5.5 10/6 10/6.5

40 40/5.5 40/6 40/6.5

70 70/5.5 70/6 70/6.5

100 100/5.5 100/6 100/6.5

130 130/5.5 130/6 130/6.5

Figure 16: 3. Simulation parameter setup

Results of the 3rd Simulation

The figures 17 and 18 show the visualized results of the

simulation.

Results for Intensity of 5.5






Results for Intensity 6






Results for Intensity of 6.5






In Figure 20 the different pathbreaking probabilities are

listed. With an intensity of 6.5 and a continuity of 40

and more the path breaks with a probability of 100%.

Compared to the intensity of 7 (figure 6) these two

intensities are very close the each other.

intensity (int)

con

tin

uit

y (

tick

s)

5.5 6 6.5

10 0% 0% 0%

40 0% 68% 100%

70 22% 100% 100%

100 93% 100% 100%

130 99% 100% 100%


The new path direction probability increases rapidly

from an intensity of 5.5 to 6 with a continuity of 100

from 6% to 98%, shown in figure 21. The highly

sensitive area can be bounded between the intensities

from 5.5 to 6.

intensity (int)

con

tin

uit

y (

tick

s)

5,5 6 6,5

10 0% 0% 0%

40 0% 0% 0%

70 0% 12% 18%

100 6% 98% 100%

130 81% 100% 100%


Also the average spread in figure 22 has a very sensitive

reaction in this parameter area. The combination

intensity of 5.5 and continuity of 10 has no valuable

effect on the spread, but the spread changes with an

increasing continuity to the direction of the forced

norm.

intensity (int)

co

nti

nu

ity (

tick

s)

5,5 6 6,5

10 1,00 0,91 0,85

40 0,92 0,48 0,42

70 0,64 0,11 0,07

100 0,25 -0,18 -0,21

130 -0,07 -0,39 -0,41

Figure 22: Average spread, calculated at the last

external impact tick

Conclusion

The aim of this research was to examine the behavior of

external influences of a path dependent hierarchical

organization with the method of computer simulation.

The basis of the simulation was the M1 model from

Petermann (2012) that simulates a path dependent

hierarchical organization. The model was enhanced to

simulate an external impact in form of continuity and

intensity which were combined and incorporated into

the M1 model. For the first simulation a parameter setup

that seemed to be valid during the implementation of the

external impact was used. With the first results multiple

questions arose and two more simulations with adjusted

parameter setups were executed. To get an overview of

the three simulations figure 23 shows an interesting

chart spread versus intensity.

Figure 23: Spread vs intensity. X-axis: intensity. Y-axis:

spread. Legend: Continuities from 10-130

To see a reaction on the system a critical intensity is

needed. An intensity of 3 and less has no effect on the

spread. The system first starts to react at the

combination of intensity 3 and continuity of 100. As

described in the third simulation, with an intensity of 5

the spread changes dramatically, but the intensive

change stops immediately at the intensity of 6 and over.

In this intensity field the external impact must last for a

defined continuity to adopt a new norm in the whole

company. The defined leadership impact of 1 concludes

that with an intensity of 5, which is the minimum value

to change the spread, the external impact needs to be

five times stronger than the leadership impact. To

clarify this, further research might show results.

Figure 24: Probability for a new path direction with a

continuity of 70. X-axis: intensity, y-axis: new path

direction probability. Legend: continuity of 70

Another interesting result of this research is the new

path direction probability. It was not in scope at the

beginning of this research, but we figured out that we

discovered an interesting system behavior at the

continuity of 70 that leads the spread to 0 with

intensities of 13 and more. This probability increases

more and more, the higher the intensity becomes. This

unexpected system behavior should be investigated

further in the future.

The next interesting point is the fact that path breaking

does not necessarily lead to a new path direction. With a

continuity of 40 and an intensity of 6.5 the path

breaking probability is 100%, but the new path direction

probability is 0%. The question that comes up here is:

does it make sense to speak about breaking the path

without actually changing the path? This might indicate

the necessity to adapt the definition of path breaking in

this context.

REFERENCES

Arthur, B. 1989. “Competing technologies, increasing returns

and lock-in by historical events.” The Economic Journal, 99 (March 1989), 116-131.

David, P. 1985. “Clio and the Economics of QWERTY.”

Economic History, Vol. 75, No. 2, 332-337

Davis, P, Eisenhardt, K. and Bingham, C. 2007. “Developing

Theory through Simulation Methods.” Academy of

Management Review,Vol. 32, No 4, 480-499.

Gilbert, N. and Troitzsch, K 2005. “Developing Theory

through Simulation Methods.” 2. ed, Berkshire: Open

University Press

Harrison, J, Lin, Z. and Carroll, G. 2007. “Simulation

Modeling In Organizational And Management Research.”

Academy of Management Review, Vol. 32, No. 4, S. 1229-1245.

Koch, J 2007: Strategie und Handlungsspielraum: Das

Konzept der strategischen Pfade. Zeitschrift Führung + Organisation, 76(5): 283-291

Sydow, J, Schreyögg, G. and Koch, J. 2009. Organizational

path dependence: Opening the black box. Adademy of Management Review, Vol. 34, No. 4, 689-709

Petermann, A., Klaußner, S., Senf, N.: Organizational Path

Dependence: The Prevalence of Positive Feedback Economics in Hierarchical Organizations, in: Troitzsch, K. G., Möhring,

M., Lotzmann, U. (Hrsg.): Proceedings 26th European

Conference on Modeling and Simulation, Koblenz 2012, 721-

730.

Vroom, V.H. 1964. Work and motivation, New York

Prof. Dr. Arne Petermann is Professor of

Management at the Berufsakademie für

Gesundheits- und Sozialwesen Saarland

(BAGSS). He is also Director of the

Institute for Management and Quality at

BAGSS. As a visiting scholar, he was

teaching organization science and scientific simulation

methods in the PhD-program at the School for Business

and Economics at Freie Universität Berlin from 2007 to

2014. He is founder and CEO of Linara GmbH, an

international HR agency specialized in the transnational

European healthcare sector. He is also founder and CEO

of 3P Projects GmbH, a consulting and executive

education company based in Berlin, Germany. His

research focuses on organization science, path-

dependence theory, entrepreneurship, and social

simulation, especially agent-based modeling. His

research results have been presented at the leading

international science conferences in his field, including

the Academy of Management and the American

Marketing Association Educators Conference where,

moreover, his work was recognized with a best paper

award. His work is published in books and peer-

reviewed journals.

Alexander Simon, MBA was born in Rheinfelden,

Germany and studied electrical engineering at the

University of Applied Sciences in St. Augustin,

Germany and Business Administration Berlin

University of Professional Studies. After first

professional experience in the area of software

development with focus on the .NET framework in

telecommunication and biotech industry, he works in

the tax advisory division of Ernst & Young since 2014.

During his MBA program at the Berlin University for

Professional Studies he became acquainted with path-

dependence theory.

Documents

Path dependence in hierarchical organizations: The ... · PDF filePath dependence in hierarchical organizations: The influence of environmental ... technological change on the TV market